Skip to main content



      Home
Home » Modeling » EMF » CSV Resource implementation for EMF
CSV Resource implementation for EMF [message #1804442] Tue, 26 March 2019 04:05 Go to next message
Eclipse UserFriend
Hi EMF community!

I have shared a few days ago an implementation of EMF Resource based on CSV format here: https://github.com/pgaufillet/emfcsv.

Among the expected advantages: the ability to open model files in spreadsheets, stable content through successive saves (and therefore good behavior with textual diff/merge tools and VCS like git), easy loading with nearly any programming technology.

Based on Apache Commons CSV, it is released under EPL 2.0.

It is an in-progress work: all questions, remarks and contributions are welcome. :-)

Pierre
Re: CSV Resource implementation for EMF [message #1804456 is a reply to message #1804442] Tue, 26 March 2019 07:25 Go to previous messageGo to next message
Eclipse UserFriend
Hi

Can you share an example so that we can see the difference. e.g. Ecore.ecore in CSV format.

Regards

Ed Willink
Re: CSV Resource implementation for EMF [message #1804482 is a reply to message #1804442] Tue, 26 March 2019 13:44 Go to previous messageGo to next message
Eclipse UserFriend
Please find attached the Ecore model converted to CSV.

Note that the goal is not to replace XMI but rather to complement it.

Pierre
  • Attachment: Ecore.csv
    (Size: 51.44KB, Downloaded 116 times)
Re: CSV Resource implementation for EMF [message #1804498 is a reply to message #1804482] Wed, 27 March 2019 04:40 Go to previous message
Eclipse UserFriend
Hi Pierre

Your goal is presumably to make spreadsheet queries easy ... I don't think you achieve that.

The model has three dimensions of iteration: instances, slots, elements. You align these with spreadsheet lines, columns, ??magic-separation?? which may make visual sense but consider:

I want to query all the names. In your organisation, the query must iterate across the lines and columns to find "name" features.

In a more pragmatic, one line per element, the query locates all the column 2's with a "name" value, then selects all the corresponding column 3's.

With each element on a different line you avoid the complexity that horrible CDATA may be in many different columns, or multi CDATA in a single cell; easier if always a separate column 3 cell per value.

With each element on a different line, you have fixed purposes for each column, and can add meta-columns such as isCDATA. Users can add their own analysis columns.

(Have you checked the Spreadsheet publications of the Epsilon team?)

Regards

Ed Willink
Previous Topic:Loading models/ metamodels from classpath resources.
Next Topic:SVBR ecore model
Goto Forum:
  


Current Time: Sat Jul 12 00:12:03 EDT 2025

Powered by FUDForum. Page generated in 0.24778 seconds
.:: Contact :: Home ::.

Powered by: FUDforum 3.0.2.
Copyright ©2001-2010 FUDforum Bulletin Board Software

Back to the top